We have compiled a list of the 10 most frequently asked questions regarding hyperspectral imaging based on our customer visits and demo tours. We’ve divided this list to 2 parts, starting with the five easiest ones and then on our next post will round out the full “Top 10” list with little more complicated questions. So stay tuned for the next installments of the series in a few days.

1 – What is a hyperspectral imaging or hyperspectral camera?

Whereas the human eye sees color of visible light in mostly three bands (red, green and blue), spectral imaging divides the spectrum into many more bands. So in comparison to a “normal” camera, an SLR or a point-and-shoot type camera, a hyperspectral camera does not record images with 3 bands only, but with hundreds of bands. As an illustration, our hyperspectral cameras dedicated for example to the visible spectral region (VIS 390 – 700 nm) record images over at least 750 bands.

Now for each of these bands, an image is created, coded with grey scale levels (as for a RGB image there will be the Red, Green and Blue components of the image). When you combine these images or bands you form a hyperspectral data cube for processing and analysis.

In short hyperspectral imaging combines digital imaging with spectroscopy – provides spectral information in each image pixel. Each pixel can then be associated with the spectral signature of the target. This result can be then used to identify, measure and locate different materials and their chemical and physical properties.

2 – Can it help to solve my problem?

It depends on the problem. The most important thing is  to know if there are spectral features that can be related to the problem:

  • where these spectral features are and
  • how it is influenced by the property which is monitored.

Besides, it is important to keep in mind that in all hyperspectral applications, the camera cannot see through the sample, as an X-Ray system would do. The penetration depth of an hyperspectral image is very low. The only possibility to see through would be to use a transmissive setup (the sample would be located between the illumination and the hyperspectral camera), but this requires that

  • the sample is not completely opaque and
  • the illumination is strong.

Once the previous is taken into account and fulfilled, qualitative or quantitative analysis can be done to characterize the samples.  In a qualitative manner the material of the sample is identified (e.g. separate samples made of PVC and ABS), whereas in a quantitative analysis the concentration of components is determined (e.g. fat content, moisture levels…)

3 – Can I process the data with Photoshop? Matlab?

Photoshop can handle RGB images and divide them into different bands/channels, but not the hyperspectral ones. So, you cannot process hyperspectral data in Photoshop.

However the raw data of Specim cameras can be imported into Matlab.

4 – How big can my samples be?

If we talk about a normal demo setup that we use in trade shows and in customer visits, we can scan samples as large as 20 x 40 x 10 cm (< 4 kg) with our LabScanner 40×20. This is the most used setup that can show the principle of a hyperspectral system.

For bigger samples, we have LabScanner 100×50, with a sample tray for 100 cm x 50 cm samples.  If you have bigger samples, please ask for a quote for a demo measurement campaign or contact us for more information.

5 – What is the price?

Of course, this is the most often-asked question. But I am not going to answer it here.

Contact us for more information, and we’ll determine together whether hyperspectral imaging is the solution you are looking for.

Mathieu Marmion holds a double MSc degree in Electrical Engineering (INP Grenoble France and NTNU Trondheim Norway) as well as a PhD in Physical Geography (Oulu Finland). For more than the past 10 years, Mathieu has been employed at Specim and has been a Technical Sales Engineer and a Sales Manager before to be currently a Lead Application Specialist.